Citation
Ferdi Kossmann, Ziniu Wu, Eugenie Lai, Nesime Tatbul, Lei Cao, Tim Kraska, and Sam Madden. 2023. Extract-Transform-Load for Video Streams. Proc. VLDB Endow. 16, 9 (May 2023), 2302–2315. https://doi.org/10.14778/3598581.3598600Rights
Copyright is held by the owner/author(s). Publication rights licensed to the VLDB Endowment. This work is licensed under the Creative Commons BY-NC-ND 4.0 International License.Collection Information
This item from the UA Faculty Publications collection is made available by the University of Arizona with support from the University of Arizona Libraries. If you have questions, please contact us at repository@u.library.arizona.edu.Abstract
Social media, self-driving cars, and traffic cameras produce videostreams at large scales and cheap cost. However, storing and queryingvideo at such scales is prohibitively expensive. We propose totreat large-scale video analytics as a data warehousing problem:Video is a format that is easy to produce but needs to be transformedinto an application-specific format that is easy to query.Analogously, we define the problem of Video Extract-Transform-Load (V-ETL). V-ETL systems need to reduce the cost of running auser-defined V-ETL job while also giving throughput guaranteesto keep up with the rate at which data is produced. We find thatno current system sufficiently fulfills both needs and therefore proposeSkyscraper, a system tailored to V-ETL. Skyscraper can executearbitrary video ingestion pipelines and adaptively tunes them toreduce cost at minimal or no quality degradation, e.g., by adjustingsampling rates and resolutions to the ingested content. Skyscrapercan hereby be provisioned with cheap on-premises compute anduses a combination of buffering and cloud bursting to deal withpeaks in workload caused by expensive processing configurations.In our experiments, we find that Skyscraper significantly reducesthe cost of V-ETL ingestion compared to adaptions of current SOTAsystems, while at the same time giving robustness guarantees thatthese systems are lacking.Note
Open access articleISSN
2150-8097Version
Final published versionae974a485f413a2113503eed53cd6c53
10.14778/3598581.3598600
Scopus Count
Collections
Except where otherwise noted, this item's license is described as Copyright is held by the owner/author(s). Publication rights licensed to the VLDB Endowment. This work is licensed under the Creative Commons BY-NC-ND 4.0 International License.